Aliso: a Journal of Systematic and Evolutionary Botany Kalinia, a New North American Genus for a Species Long Misplaced in Eragrostis (poaceae, Chloridoideae) Kalinia, a New North American Genus for a Species Long Misplaced in Eragrostis (poaceae, Chloridoideae)

Eragrostis obtusiflora (Poaceae, Chloridoideae), a species distributed from the southwestern United States to central Mexico, has long been recognized for exhibiting morphological and anatomical features atypical of Eragrostis. Phylogenetic analyses of nuclear internal transcribed spacer sequences and plastid sequences demonstrate that E. obtusiflora should be excluded from Eragrostis (Eragrostideae) and instead be placed in Cynodonteae, although its position within this tribe was unresolved. Additional data, including anatomical and micromorphological characters, suggest a close relationship with Distichlis. However, differences in spikelet and rhizome characters prevent its inclusion in Distichlis. Therefore, the species is transferred to a newly described genus, Kalinia, as K. obtusiflora.


INTRODUCTION
In the fall of 2001, as part of a systematic study of Distichlis Raf. and relatives (Poaceae, Chloridoideae; Bell and Columbus 2008), the first two authors collected D. spicata (L.) Greene from numerous populations in the southwestern USA and Mexico.At Willcox Playa (Cochise County, Arizona, USA) and Salinas de Hidalgo (San Luis Potosı ´, Mexico) few plants of D. spicata were found to be in flower.We noted many non-flowering plants that were similar to D. spicata in overall appearance and habit but differed by having markedly pungent-tipped leaves, in contrast to the pointed but not pungent leaf blades of D. spicata.Subsequent analyses of DNA sequences of the nuclear internal transcribed spacer (ITS; nrDNA) and trnL-F (cpDNA) regions revealed that sequences of pungent-leaved plants from the distant localities were nearly identical, yet differed from other sequences of D. spicata, and, indeed, all other species of Distichlis.
In fall 2003 we revisited Willcox Playa and found the pungent-leaved plants in flower.The flowers were hermaphroditic in contrast to the unisexual flowers of dioecious Distichlis.In addition, the 3-nerved lemmas, versus 5-13 in Distichlis, helped us identify the plants as Eragrostis obtusiflora (E.Fourn.)Scribn.(Eragrostideae), a species distributed from the southwestern USA (Arizona, New Mexico) to central Mexico in saline/alkaline soils of inland playas (Pleistocene lake beds; Rosen 1994;Briere 2000).Distichlis spicata is often sympatric with E. obtusiflora, and, as pointed out by McVaugh (1983), non-flowering plants are easily confused.Both species are rhizomatous and have rigid, often conspicuously orthodistichous leaf blades.
In morphological and anatomical studies of Eragrostis Wolf, a large, worldwide genus of ca.400 species (Clayton et al. 2006 onwards), Van den Borre and Watson (1994) and Go ´mez Sa ´nchez and Koch (1998) reported that E. obtusiflora is anomalous within Eragrostis.In particular, E. obtusiflora differs from other species of Eragrostis in having pungenttipped leaves, a 3-nerved upper glume, bundle sheaths that are not interrupted, bulliforms consisting of two large cells and associated with girders of colorless cells to the abaxial epidermis, and papillae present on intercostal long cells.In contrast, similarities between E. obtusiflora and members of Monanthochloinae (Clayton and Renvoize 1986), including Distichlis, have been noted by a number of authors, specifically in overall growth habit, presence of well developed rhizomes, and occurrence in saline and/or alkaline habitats (Ogden 1897;Scribner 1897;Go ´mez Sa ´nchez and Koch 1998;Bell and Columbus 2008).In addition, E. obtusiflora shares two characteristics with species of Distichlis: bicellular microhairs where the basal cell is sunken into the mesophyll (also present in other halophytic grasses) and, in the intercostal zones of abaxial leaf blade surfaces, microhairs are surrounded and sometimes overarched by groups of four papillae (Bell and Columbus 2008).
The Grass Phylogeny Working Group II [GPWG II] (2012) conducted a family-wide analysis based upon three plastid loci and using a broad sample of Chloridoideae that included E. obtusiflora.Eragrostis obtusiflora resolved in the Cynodonteae clade, distant from the Eragrostideae clade where all other sampled Eragrostis species resolved.Although sister to Distichlis in the tree, the relationship had low statistical support.
Based on Ogden (1897), Van den Borre and Watson (1994), Go ´mez Sa ´nchez and Koch (1998), and GPWG II (2012), there is growing evidence that E. obtusiflora is misplaced in Eragrostis.We test this hypothesis by analyzing DNA sequences of additional loci from samples of E. obtusiflora and many representatives of Chloridoideae, including Distichlis and a large number of Eragrostis species.In addition, the abaxial surfaces of leaf blades and lemmas and blade transectional anatomy of E. obtusiflora are further examined and compared to other chloridoid grasses.

Sampling
All specimens used in this study are listed in Appendix 1. Specimens of E. obtusiflora from Arizona, USA, (Bell 295) and, in Mexico, San Luis Potosı ´ (Bell 251) and Jalisco (Bell 318) were dried in silica gel for DNA extraction and sequencing.ITS and trnL-F sequences were added to the data sets from Columbus et al. (2007) which include a broad sample of Chloridoideae.The rps16 (cpDNA) region of Bell 251 was sequenced and added to the data set from Ingram and Doyle (2007).This data set contains a large sample of Eragrostis species representing both subgenera (Van den Borre and Watson 1994) and most of the major morphological groups in the genus based upon spikelet disarticulation type.
Specimens from across the geographical range of E. obtusiflora were examined for leaf blade transectional anatomy (Bell 295,305,318) and for micromorphology of the abaxial surfaces of leaf blades (Bell 295,310,318) and lemmas (Bell 239,295,305).As described in Bell and Columbus (2008), segments of living leaf blades were liquid-preserved in the field for anatomical study.

DNA Sequencing and Analysis
Sequences of ITS, trnL-F, and rps16 from E. obtusiflora were obtained following the methods in Columbus et al. (2007) and Ingram and Doyle (2004).The sequences were manually aligned with the data sets of Columbus et al. (2007) and Ingram and Doyle (2004) and analyses of these data were carried out as described in those studies.GenBank accession numbers for newly generated sequences are provided in Appendix 1.We also performed a Kishino-Hasegawa test (Kishino and Hasegawa 1989) and a Templeton test (Templeton 1983) using the combined ITS and trnL-F data set in PAUP vers.4.0b (Swofford 2002) to compare the length of the most parsimonious trees with the length obtained when E. obtusiflora was constrained as a member of Eragrostideae (sensu Columbus et al. 2007).

Micromorphology
The abaxial surfaces of fully expanded leaf blades and mature lemmas were examined using scanning electron microscopy following the methods of Bell and Columbus (2008).

Leaf Blade Transectional Anatomy
Preparation of blade transections followed the methods described in Columbus (1996) and Bell and Columbus (2008).Permanent microscope slides are deposited at RSA. Descriptive terminology for leaf anatomy and micromorphology follows Ellis (1976Ellis ( , 1979)).

Analyses of DNA Sequences
In maximum parsimony analyses of the separate (trees not shown) and combined (Fig. 1) data sets of ITS and trnL-F, E. obtusiflora resolved within Cynodonteae, apart from the other species of Eragrostis.However, due to limited resolution and clade support, the phylogenetic position of E. obtusiflora within Cynodonteae is uncertain.In maximum parsimony analysis of rps16, E. obtusiflora likewise was not placed with other Eragrostis species and instead resolved in a clade among members of Cynodonteae (Fig. 2).Descriptive statistics for the analyses are given in Table 1.Results of a Kishino-Hasegawa test indicated that tree length was significantly longer when E. obtusiflora was constrained as part of Eragrostideae (length difference 5 42, p 5 0.001; see Fig. 1).A Templeton test provided almost identical results.

Micromorphology
As seen in Fig. 3, costal and intercostal zones of the abaxial leaf blade surface are distinct in E. obtusiflora.Macrohairs and prickle hairs were not observed.Microhairs, papillae, and stomata are only present in the intercostal zone.The shape of intercostal long cells is obscured by the numerous large rounded papillae.There are two rows of stomata each restricted to one edge of the intercostal zone; subsidiary cells are low-dome shaped.Intercostal short cells generally occur singly; they are vertically oblong or reniform.Microhairs are distributed along the middle of the intercostal zone and are each surrounded and slightly overarched by groups of papillae.From anatomical studies we know that the microhairs are bicellular (Fig. 7); however, in the surface view only the distal cell is visible and frequently is collapsed, indicating the cell wall is thin.
The costal zone is raised and made up of seven to nine files of cells.The center file contains short cells that alternate in shape between round and more-or-less square with smooth walls.On both sides of the center file there are files of narrow long cells with undulating margins intermixed with short cells of both kinds.
Overall, the abaxial surfaces of blades and lemmas (Fig. 4) of E. obtusiflora are quite different from each other.The blade surface has a regular pattern of alternating costal and intercostal zones determined by the numerous veins.Lemmas have three nerves and the costal and broad intercostal zones are not as distinct.Long cells have deeply undulating walls.Short cells are oval to reniform in shape and occur singly between long cells.Lemmas of E. obtusiflora have few papillae, Bootstrap values $50 are given above the branches.Phylogenetic analysis methods are described in Ingram andDoyle (2004, 2007).
prickle hairs, stomata, and microhairs (Fig. 4).Stomata on lemmas have a similar shape to those on the leaf blades with lowdome-shaped subsidiary cells.Papillae are found adjacent to and sometimes overarching both stomata and microhairs.Apically pointing prickle hairs are distributed in the costal and intercostal zones.Microhairs on lemmas are the chloridoid bicellular type.
The observations reported in this study are generally in agreement with those of Go ´mez Sa ´nchez and Koch (1998) with the exception that their illustration of the abaxial blade surface shows five files of cells in the costal zone (versus 7-9 in the present study) and they do not show any round short cells in these files.Go ´mez Sa ´nchez and Koch (1998) did note rows of stomata with overarching papillae in the intercostal zones.Liu et al. (2010) did not find stomata on lemmas of the five Eragrostis species that were included in their study.

Leaf Blade Anatomy
Transectional leaf anatomy of E. obtusiflora is distinctive with large ridges of colorless cells on the adaxial side of all vascular bundles except those at the margins (Fig. 5, 6).The outline of the blade forms a broad, continuously curved U.There are deep, narrow furrows between all vascular bundles on the adaxial side and shallow furrows on the abaxial side.The median vascular bundle is not distinct as a midrib and there is no keel.The blade has 19-25 vascular bundles with five 1 st order bundles regularly arranged with two or three 2 nd order bundles between the 1 st order and a single 3 rd order bundle at each margin.
The shape of all vascular bundles is elliptical or slightly pointed adaxially.The vascular bundles have a double bundle sheath, the outer of which is even in outline and contains chloroplasts that are centripetally arranged within the cells, features that are predictive of the NAD-ME type of C 4 photosynthesis (Prendergast and Hattersley 1987).In the 1 st order vascular bundles, parenchyma cells adjoin the inner sheath.Metaxylem is slightly narrower than the outer sheath cells and has thickened walls.Bundle sheaths are complete; on the abaxial side some outer sheath cells lack chloroplasts.Adaxial bundle sheath extensions consist of large, thin-walled, colorless cells that are separated from the epidermis by a narrow sclerenchyma strand.Abaxial sclerenchyma forms small girders.At each margin there is a small, slightly pointed sclerenchyma cap.The mesophyll is radiate, composed of tabular cells, and separated from adjacent vascular bundles by bi-or multiseriate columns of colorless cells of irregular size and shape.At the base of the furrows there are bulliform cells associated with the colorless cells.
There are numerous papillae and/or prickle hairs on the adaxial surface and, consistent with the micromorphological study, fewer papillae on the abaxial surface between vascular bundles.Dumbbell or flask-shaped bicellular microhairs are present on both surfaces; the basal cell is partially sunken below the epidermis into colorless or mesophyll cells (Fig. 7).
The results of the present study are in general agreement with previous work, with the exception that Go ´mez Sa ´nchez and Koch (1998) did not distinguish 2 nd order and 3 rd order vascular bundles in blades of E. obtusiflora.However, distinct metaxylem is evident in the 1 st order vascular bundles (Ellis 1976) as noted by Ogden (1897) and Bell and Columbus (2008).

DISCUSSION
Eragrostis obtusiflora has long been positioned in Eragrostis because of its multiple florets per spikelet and 3-nerved, awnless, and glabrous lemmas.However, based on other characteristics, including leaf anatomy, taxonomists have found it to be a puzzling species with enigmatic affinities (Ogden 1897;Scribner 1897;Hitchcock 1951;McVaugh 1983; Van den Borre and Watson 1994; Go ´mez Sa ´nchez and Koch 1998; Caldero ´n de Rzedowski and Rzedowski 2001).Most of these researchers have noted similarities in vegetative appearance and in habitat requirements between E. obtusiflora and members of Monanthochloinae, especially D. spicata.Distichlis spicata and E. obtusiflora frequently co-occur in saline/alkaline habitats, and both species are strongly rhizomatous.However, rhizomes of E. obtusiflora are thicker (.5 mm) and generally golden yellow with distinctive scale scars while rhizomes of D. spicata are more slender (,4 mm) and brownish with persistent scale leaves.In addition, lemma nerve number (3 in E. obtusiflora vs. 7-13 in D. spicata) and floral sexuality (hermaphroditic in E. obtusiflora vs. unisexual in D. spicata) clearly distinguish these species.
Analyses of molecular data from the nuclear and chloroplast genomes confirm that E. obtusiflora has been misplaced taxonomically (GPWG II 2012; Fig. 1, 2).It does not form a clade with other species of Eragrostis in tribe Eragrostideae.When E. obtusiflora is constrained as part of Eragrostideae (in the analysis of combined ITS and trnL-F, Fig. 1), tree length is significantly longer (Templeton 1983;Kishino and Hasegawa 1989).Rather, it resolves in Cynodonteae.In both GPWG II ( 2012) and the present study, it resolves with the same group of genera but with a lack of statistical support making its exact relationship to these genera uncertain based upon these data and analyses.
In addition, E. obtusiflora has stomata on its lemmas (Fig. 4), a feature not observed in a small sample of Eragrostis species (Liu et al. 2010).In Chloridoideae, stomata on lemmas have been reported in 25 (of ca.140) genera including Distichlis and other related members of Cynodonteae (Columbus 1996;Bell and Columbus 2008;Liu et al. 2010).
Leaf blade anatomy provides additional evidence of the unique nature of E. obtusiflora.Four previous studies have presented data on blade transectional anatomy of this species (Ogden 1897;Van den Borre and Watson 1994;Go ´mez Sa ´nchez and Koch 1998;Bell and Columbus 2008).The conclusion of these studies is that anatomy of E. obtusiflora differs from that of other Eragrostis species by the presence of papillae in intercostal zones of the abaxial surface and the lack of interruption of the bundle sheath by sclerenchyma (shared only with E. pergracilis S.T.Blake).The anatomy of E. obtusiflora differs from Distichlis in the presence of bundle sheath extensions of colorless cells (absent in Distichlis) and the wide diameter of metaxylem cells (narrow in Distichlis) (Bell and Columbus 2008).However, the presence of dumbbell or flask-shaped bicellular microhairs with a portion of the basal cell sunken below the epidermis is a character that is shared by E. obtusiflora and all members of Distichlis, supporting a possible close relationship between these taxa (Fig. 7, 8; Bell and Columbus 2008).
It is clear from analyses of DNA sequence data and from anatomical and morphological evidence that E. obtusiflora does not belong in Eragrostis and, in fact, is placed in Cynodonteae, but there is no evidence for placing it in any other existing genus.Distichlis may be most closely related, but it differs most evidently by its 5-13-nerved lemmas and unisexual flowers.Therefore, we propose a new genus, Kalinia, for this species.Robust perennial with long, glabrous stolons and thick, tan to golden rhizomes with imbricate scales dehiscing to form distinctive scars; ligule a line of short hairs, leaf blades rigid, conspicuously orthodistichous, with markedly pungent tips; inflorescence a narrow to open panicle; spikelets with 5[4?]-12 hermaphroditic florets, the distal floret occasionally reduced, lemmas 3-nerved, glabrous, with an obtuse to slightly pointed tip, sometimes erose, fringed, or mucronate from the central nerve.
Etymology.-Thegeneric name is derived from the Arabic root of alkali, al qaliy (ashes of saltwort), in recognition of the habitat preference of this plant.
Type locality.-Inthe Brizopyrum obtusiflora protologue, Fournier (1886) indicated ''[i]n ora occidentali (E ´my in meo herbario).''''Orizaba'' is written on the type specimen.Fournier (1878) noted that some specimens collected by members of the French Scientific Commission to Mexico, 1865-1866, did not have detailed collection data.Such specimens were attributed to Captain E ´my and to the site of the military encampment, Orizaba, in the state of Veracruz (Hemsley 1882(Hemsley -1886)).Fournier mentioned Acapulco (Guerrero) and Mazatla ´n (Sinaloa) as coastal areas where collections were made by members of the commission; however, K. obtusiflora has not been collected from the Pacific (or Gulf) coast of Mexico, including Veracruz.We are not aware of suitable habitat of K. obtusiflora near Orizaba.In the region it is known from collections at Laguna Atotonilco (Jalisco), Laguna Cuitzeo (Michoaca ´n), and Lago de Texcoco (Me ´xico).
Distribution and habitat.-Kalinaobtusiflora occurs in Arizona and New Mexico (USA) and in the Mexican states of Chihuahua, Coahuila, Durango, Guanajuato, Jalisco, Me ´xico, Michoaca ´n, and San Luis Potosı ´(Appendix 1; Fig. 10).It is frequently sympatric with D. spicata.A large number of specimens are from Willcox Playa in southern Arizona, and multiple vouchers represent the area around the town of Playas, New Mexico, as well as northern Chihuahua and the

Table 1 .
Summary information for the data sets and results of the analyses.PIC 5 parsimony informative characters, CI 5 consistency index, RI 5 retention index.